ggerganov/llama.cpp

Pull Requests Commits

unify scalar+vector and fix reduce function

0cc4m committed 35 days ago

5ec6569e

Revert "move kv shmem staging to function"

0cc4m committed 35 days ago

e880cb2e

move kv shmem staging to function

0cc4m committed 35 days ago

0349025d

generic reductions

0cc4m committed 35 days ago

2c623bfa

fix slang issues

0cc4m committed 35 days ago

e1b40fa5

vulkan: port Flash Attention shader to Slang

0cc4m committed 35 days ago

a4ac1d90

convert : fix/suppress pyright errors (#20442)

danbev committed 35 days ago

Verified 983df142

llama : disable graph reuse with pipeline parallelism (#20463)

ggerganov committed 35 days ago

Verified 57819b8d

vendor : update cpp-httplib to 0.37.1 (#20390)

cabelo committed 36 days ago

Verified 557fe2d9

tests : use `reasoning` instead of `reasoning_budget` in server tests (#20432)

pwilkin committed 36 days ago

Verified 0e810413

test-backend-ops: allow loading tests from file and parsing model operators into file (#19896)

0cc4m committed 36 days ago

Verified 128142fe

common : update completion executables list [no ci] (#19934)

danbev committed 36 days ago

Verified 6de1bc63

grammar: Fix grammar root symbol check (#19761)

AsbjornOlling committed 36 days ago

Verified 0a10c34d

vulkan: add GATED_DELTA_NET op support (#20334)

ProgenyAlpha committed 36 days ago

Verified deee2386

convert : better mtp check and fix return [no ci] (#20419)

CISC committed 36 days ago

Verified c3e3f9e5

vulkan: fix SSM_CONV PP scaling with large ubatch sizes (#20379)

ProgenyAlpha committed 36 days ago

Verified 40c550d4

New conversations now auto-select the first loaded model (#20403)

ServeurpersoCom committed 36 days ago

Verified de190154

ggml-virtgpu: Fix some build commands (#20341)

yomaytk committed 36 days ago

Verified 05039967

metal : avoid divisions in bin kernel (#20426)

ggerganov committed 36 days ago

Verified e4cff095

ci: Setup self-hosted CI for Intel Linux Vulkan backend (#20154)

rillomas committed 36 days ago

Verified 4cc6eb15

vulkan: fix l2_norm epsilon handling (#20350)

jeffbolznv committed 36 days ago

Verified 246ffc4b

vulkan: fix OOB check in flash_attn_mask_opt (#20296)

jeffbolznv committed 36 days ago

Verified aa429cf5

vulkan: Fix ErrorOutOfHostMemory on Intel GPU when loading large models with --no-mmap (#20059)

rillomas committed 36 days ago

Verified 5866e3bb

opencl: use larger workgroup size for get_rows (#20316)

lhez committed 36 days ago

Verified 0516e04b

opencl: add cumsum op (#18981)

shaofeiqi committed 36 days ago

Verified 3d9ab225

hip: compile debug builds with -O2 on hip to avoid a compiler bug (#20392)

IMbackK committed 36 days ago

Verified d63aa398

common/parser: add GigaChatV3/3.1 models support (#19931)

Mishusha committed 36 days ago

Verified a8304b4d

model : add support for Phi4ForCausalLMV (#20168)

dranger003 committed 36 days ago

Verified fdb17643

graph : add optional scale parameter to build_lora_mm [no ci] (#20427)

richarddd committed 36 days ago

Verified 1eea6a29

common : fix --n-cpu-moe, --cpu-moe for models with fused gate + up (#20416)

ddh0 committed 36 days ago

Verified 4a748b8f

Older